The SkipList newest entered elements are on its front:

So this is broken:

Delayed post-proc uses lower_bound, which is the cleanest and preferred way for every implementation.

Yet another bug that highlights the importance of reworking the post-processing facility in the datahandlingmodel.

All the 3 ways are depending on lot of assumptions, and all 3 breaks in different ways.
Yet another highlight, that sometimes defaults and generics are there, to specialize them to use-cases. One example: The BinarySearchQueueModel.
Or the frame specific processors.

But the Default reception and TaskBased processors must be specialized, and not trying to cover every corner cases and conditions.